Drop-in OpenClaw / NemoClaw agent configuration for the ibl.ai K-12 solution segment — one School Assistant orchestrator plus 12 specialist subagents, packaged so a NemoClaw host can import the whole roster with zero conversion.
This directory contains a complete, drop-in NemoClaw/OpenClaw agent configuration for the K-12 education segment. It defines one parent "School Assistant" agent and twelve specialist subagents covering the full range of K-12 workflows from adaptive tutoring to district administration.
| Agent ID | Name | Role |
|---|---|---|
k-12-assistant |
School Assistant | Segment entry point; greets users, interprets intent, delegates to specialists |
tutoring-agent |
Tutor | Adaptive academic support in math, reading, and science |
lesson-planning-agent |
Lesson Planner | Standards-aligned lesson and unit plan creation |
assessment-agent |
Assessment Builder | Quiz generation, rubric creation, and auto-grading |
writing-feedback-agent |
Writing Coach | Student writing review and revision guidance |
special-education-agent |
Special Education Support | IEP, 504, and accommodation support |
content-creation-agent |
Content Creator | Worksheets, presentations, and classroom activities |
student-safety-agent |
Safety Monitor | Content moderation and student safety guardrails |
family-communication-agent |
Family Communicator | Parent updates, newsletters, and multilingual outreach |
curriculum-alignment-agent |
Curriculum Aligner | Standards mapping and curriculum gap analysis |
professional-development-agent |
PD Coach | Teacher professional development and coaching |
research-agent |
Research Guide | Student research support and citation guidance |
administration-agent |
School Administrator | Scheduling, reporting, and district operations |
The parent agent (k-12-assistant) is the default entry point. It accepts requests from teachers, students, families, and staff, then delegates to the appropriate specialist subagent via sessions_spawn.
k-12/
├── README.md <- this file
├── openclaw.json <- NemoClaw gateway configuration
├── .config-hash <- sha256 of openclaw.json
├── .env.example <- sample API credential variables (copy to ~/.openclaw/.env)
├── workspace/
│ └── .gitkeep <- shared writable workspace (empty at deploy time)
├── skills/
│ └── <platform>/
│ └── SKILL.md <- tool-skill definition for a third-party integration
└── agents/
└── <agent-id>/
└── agent/
├── auth-profiles.json
├── IDENTITY.md
├── SOUL.md
├── TOOLS.md
├── AGENTS.md <- parent agent only
├── MEMORY.md <- optional: seed long-term facts (regulation/standards-heavy agents)
├── HEARTBEAT.md <- optional: periodic monitoring checklist (schedule-driven agents only)
├── BOOTSTRAP.md <- optional: one-time first-run setup steps (consumed after first use)
└── USER.md <- optional: user/environment-specific context (rarely needed)
Follow these steps to import this configuration into a NemoClaw sandbox.
-
Copy this directory into the sandbox. Copy the entire contents of this directory into
/sandbox/.openclaw/on the NemoClaw host. After copying, the path/sandbox/.openclaw/openclaw.jsonmust exist.git clone https://github.com/iblai/k-12-agents.git
cp -r k-12-agents/. /sandbox/.openclaw/
2. **Set ownership on the agents directory.**
Ensure the `agents/` subdirectories and workspace are owned by the sandbox user (typically `sandbox`):
```bash
chown -R sandbox:sandbox /sandbox/.openclaw/agents /sandbox/.openclaw/workspace
-
Replace the sample API credentials. Every agent directory contains
agent/auth-profiles.jsonwith placeholder values. Replace theapiKeyvalue in each file with a real Anthropic API key before starting the gateway:# Example: update one agent nano /sandbox/.openclaw/agents/k-12-assistant/agent/auth-profiles.jsonThe placeholder value is:
sk-ant-api03-SAMPLE-PLACEHOLDER-NOT-A-REAL-KEY-0000000000000000000000000000000000000000 -
Set up third-party platform credentials. Copy
.env.exampleto~/.openclaw/.envand replace every placeholder value with a real API key or OAuth credential for the platforms your deployment will use:cp /sandbox/.openclaw/.env.example ~/.openclaw/.env nano ~/.openclaw/.env
.env.exampleis committed to the repository as a sample template representing the OpenClaw daemon env file~/.openclaw/.env; all its values are non-functional placeholders. The real~/.openclaw/.envshould never be committed. See the Tool Skills section for the full list of platforms and the variables each one requires. -
Recompute the configuration hash. If you modify
openclaw.jsonafter copying, recompute the hash so the gateway accepts the configuration:cd /sandbox/.openclaw shasum -a 256 openclaw.json > .config-hash
On Linux use
sha256suminstead ofshasum -a 256. -
Reload the OpenClaw gateway. Restart or reload the OpenClaw service to pick up the new configuration:
systemctl reload openclaw # or openclaw reload
The k-12-assistant agent is marked "default": true in openclaw.json and will be the agent that receives incoming sessions unless a specific agent ID is requested.
The skills/ directory contains one subdirectory per third-party platform, each with a SKILL.md file. Each file describes the platform's API surface, the operations agents can perform, and the credential variables required. All skills draw credentials from ~/.openclaw/.env on the OpenClaw daemon host — .env.example is the committed sample template with non-functional placeholder values that you copy and fill in during deployment (see step 4 of the import instructions above).
| Skill directory | Description |
|---|---|
canvas/ |
Canvas LMS (Instructure) — read and write courses, modules, assignments, rubrics, grades, and submissions for K-12 classroom workflows. |
classdojo/ |
ClassDojo — post classroom story updates and send direct messages to authorized guardians for K-12 family engagement and communication. |
ebsco/ |
EBSCO school databases (EBSCOhost, Explora) — search peer-reviewed articles, reference materials, and grade-leveled nonfiction from school-licensed academic databases. |
edulastic/ |
Edulastic — generate, import, and manage standards-aligned assessments; retrieve performance data and class mastery reports for K-12 assessment workflows. |
frontline/ |
Frontline Education suite — access HR, absence management, professional development, and special education data across Frontline's K-12 platform products. |
google-classroom/ |
Google Classroom — read and write courses, coursework, materials, and student submissions via the Google Classroom API for K-12 workflows. |
google-workspace-edu/ |
Google Workspace for Education — read and write files in Google Drive, create and edit Docs and Slides, and manage folder structure for K-12 content workflows. |
iready/ |
iReady (Curriculum Associates) — retrieve student diagnostic placement levels and domain scores to align tutoring and instructional content to each student's instructional zone. |
khan-academy/ |
Khan Academy — retrieve student skill mastery, exercise progress, and course completion data to personalize tutoring and avoid re-teaching mastered content. |
nwea-map/ |
NWEA MAP Growth — retrieve student RIT scores, instructional area recommendations, and growth projections to calibrate tutoring and instructional support. |
parentsquare/ |
ParentSquare — compose, schedule, and send school-to-family communications (mass messages, direct messages, announcements) and read delivery analytics. |
powerschool/ |
PowerSchool SIS — read and write student records, enrollment, attendance, grades, and demographic data for K-12 school and district operations. |
- All agents follow FERPA and COPPA guidelines: student PII is never transmitted outside the sandbox.
- The
student-safety-agentfollows CIPA requirements for content filtering in school environments. - The
special-education-agentfollows IDEA 2004 and Section 504 of the Rehabilitation Act. - All sandbox processes run as the non-root
sandboxuser with seccomp and landlock enabled.
Released under the MIT License.